Anticipatory Freight Consolidation in Intermodal Networks using Approximate Dynamic Programming



We study the planning problem of transporting freights that have different characteristics, in an intermodal network, over a multi-period horizon. Although freights and their characteristics become known gradually over time, there is probabilistic knowledge about them. The goal is to use this knowledge and balance current and future costs, in order to minimize the total costs over the planning horizon. To model and optimize this tradeoff, we propose a Markov Decision Process (MDP) model and an Approximate Dynamic Programming (ADP) approach. We show the benefits and difficulties of our look-ahead approach, under different problem settings and compared to other planning approaches.

Registration to Remy Spliet, is required for availability of lunch.